minimum track set, probability of occurrence of top event．Structural significance, probability significance and critical significance of fault tree．
Fault tree chart of scraper conveyor in fully mechanized mining face is established and the qualitative analysis and quantitative analysis of the fault tree are carried out．According to analyzing results, some countermeasures to prevent accident from happening are put forward．
4．Analysis of environment factors
In man-machine-environment system, environment is a fundamental factor influencing system safety．It is the soil where unsafe behavior of workers and unsafe state of machinery are produced．
4．1．Grey correlative analysis between, mining accidents and environmental-factors
On the basis of analyzing qualitatively the influence of main environment factors to coal mining safety, the relationship between mining accidents and environment factors such as temperature, humidity, illumination, yawp and gas is analyzed with the method of grey correlative analysis[5]．
First, make the original data standardized．Secondly, calculate the correlative coefficient．Finally, work out grey correlative degree．The calculation result indicates that grey correlative degree between mining accidents and gas is the largest．Therefore, gas is preponderant factor
influencing mining accidents in Xinzhuang coal mine．
4．2．Fuzzy clustering analysis of mining laces environment
Clustering analysis is a method to classify objects according to some conditions or attributes of the objects．It is widely used in production and daily life． Environmental factors in mining faces are complex and uncertain．So how to classify is fuzzy．
First, make the original data standardized．Secondly, establish fuzzy relationship．Finally, classify objects with the method of fuzzy clustering analysis[6]．All the mining faces in Xinzhuang coal mine are classified with this method．The result indicates that the environment of fully mechanized mining face is the best．The result also tells us how to improve the coal mining environment and where is the emphasis to strengthen management．This helps to decrease environmental influence to mining accidents and to improve mining safety．
4．3．Countermeasures to improve underground environment
After analyzing qualitatively and quantitatively the influence of underground environmental factors to mining accidents, we can distinguish the primary from the secondary．We can take some measures such as strengthening ventilation, installing air conditioner, eliminating noisy devices etc to improve underground environment so as to improve mining safety and ensure workers’ safety and sustained efficient production．
5．Grey entropy, synthetic evaluation on safety of mining faces
5．1．Grey entropy
Suppose a finite discrete sequence X={xi | i=1,2,……,n},grey entropy of sequence X is defined by
If xi=0,let xi ln xi=0
H(X)?-?xi ln xi (2)
i?1n5．2．Balanced degree
Grey entropy is the measure to balance degree of elements in discrete sequence X in terms of grey entropy increase theorem[7]．The greater grey entropy is the more balanced the discrete sequence is．For a discrete sequence with n elements the maximum entropy is a constant ln n which is only relevant with the number, of elements in sequence when all elements are equal．Thus, balanced degree of sequence can be defined by
B=H(X)/Hm (3)
Where Hm is the maximum grey entropy．
5．3．Grey correlative degree
Grey correlative degree is the measure to adjacent degree between the reference sequence and relative sequence．It can be calculated by correlative coefficients[8]．The correlative coefficient can be calculated by the following
equation．
L（?ik）minminv-v??maxmaxv-viv-vik??maxmaxv-vikikk?k?kikik?k?kik(4)
Where ρ is distinguished coefficient, it is in the range from 0 to 1．Usually it is 0．5．Grey correlative degree can be expressed as．
1nroi??L（ik） (5)
nk?1Where roi is grey correlative degree between the reference sequence and the
ith relative sequence．
5．4．Model of multi-level grey entropy synthetic evaluation
If the evaluation indices lie in different levels．We should apply the model of multi-level grey entropy synthetic evaluation．Multi-level evaluation model is similar to single level evaluation model．The second-order evaluation results from the first-order evaluation matrix．The evaluation result matrix can be obtained by weight vector and the evaluation matrix in view of the significance difference of indices in the first level．It can be expressed by
W2 W=A·
where
(6)
W2 is the balanced adjacent degree matrix．
